2009年-BIS国际清算银行_Financial_Crises_and_Economic_Activity_37页_451kb

Summary of "Financial Crises and Economic Activity"

Core Content

This paper by Stephen G Cecchetti, Marion Kohler, and Christian Upper analyzes the economic impact of 40 systemic banking crises from 1980 to 2007, with a focus on the current global financial crisis of 2008. The study explores how financial crises affect real economic activity, particularly output, and identifies the factors that influence the severity and duration of these effects.

Main Findings

• Most Crises Cause Output Contraction: The majority of systemic banking crises in the dataset coincide with sharp contractions in output, from which recovery takes several years.
• Current Crisis is Unique: The 2008 financial crisis is unlike any other in the dataset, suggesting that past averages may not be reliable for predicting its outcomes.
• Output Losses Vary by Context: Output losses are higher when crises occur alongside currency crises or during periods of low growth. However, when accompanied by sovereign debt defaults, the output losses are less severe.
• Long-Term Output Effects: Systemic banking crises tend to have lasting negative effects on output, even in cases where trend growth improves post-crisis.
• Recovery Timeframe: The paper estimates that some of the main economies affected by the crisis may regain pre-crisis output levels by the second half of 2010, though the confidence interval is wide.

Key Economic Channels

The paper outlines several mechanisms through which financial crises impact the real economy:

• Funding Costs: Financial stress increases borrowing costs, as interest rates and spreads rise, and equity prices fall. This reduces investment and corporate profits.
• Credit Availability: Tightened lending standards and reduced credit supply, both through traditional channels and securitisation, lead to a contraction in economic activity.
• Risk Aversion: Increased risk aversion drives up risk premia and results in a "flight to quality," further constraining credit and investment.
• Net Worth of Firms and Households: Declines in asset prices reduce the net worth of both firms and households, increasing adverse selection and moral hazard issues.
• Exchange Rates: Currency depreciation can stimulate trade, but widespread mismatches in balance sheets can negate these benefits.
• Confidence: A drop in consumer, business, and investor confidence leads to reduced spending and investment, exacerbating economic downturns.

Methodology and Data

• Crisis Definition: A systemic banking crisis is defined as events involving a large number of defaults, difficulty in repaying contracts, and significant non-performing loans.
• Data Source: The study uses the crisis resolution database from Laeven and Valencia (2008), which includes 40 crises across 35 countries.
• Measurement of Costs: Output costs are used as a measure of real economic losses, defined as the period when output is below its pre-crisis level, the length of recovery, and the depth of the contraction.
• Conditional Analysis: The authors use conditional models to explore the determinants of output losses, focusing on initial conditions, financial structure, development level, policy responses, and external conditions.

Crisis Characteristics

• The median length of a crisis-related contraction is 8.5 quarters.
• The median depth of the contraction is 6.6% of pre-crisis GDP.
• The median cumulative loss relative to the pre-crisis GDP peak is 9.2%.
• Some crises, like the one in Bulgaria, had extremely severe effects, with GDP falling by 42% and taking almost seven years to recover.

Conclusion

The paper emphasizes the importance of distinguishing between normal and crisis conditions in economic modeling. It concludes that financial crises have significant and often lasting impacts on economic activity, and that understanding these impacts requires a nuanced, empirical approach rather than relying on historical averages. The current crisis is seen as an outlier, with unique characteristics that may require different policy responses than in the past.